The metabolite α-ketoglutarate induces AIM2-dependent PANoptosis through demethylase TET2

Abstract

While α-ketoglutarate (α-KG) has traditionally been viewed as an anti-inflammatory metabolite, we uncover its paradoxical role in driving pathological inflammation during sepsis. This study reveals that α-KG, a tricarboxylic acid cycle (TCA) intermediate elevated in septic patients, drives inflammatory macrophage death through absent in melanoma 2 (AIM2) -PANoptosome activation. Using both clinical samples and experimental models, we demonstrate that the cell-permeable derivative dimethyl-α-ketoglutarate (DM-α-KG) exacerbates lipopolysaccharide (LPS)-induced tissue injury and cell death, whereas isocitrate dehydrogenase (IDH1) inhibition (IDH-305) or genetic ablation reduces α-KG levels and confers protection. Mechanistically, α-KG enhances the dioxygenase activity of Ten-eleven translocation 2 (TET2), promoting its binding to the AIM2 promoter, reducing methylation, and increasing AIM2 expression, thereby triggering PANoptosome assembly. The pathophysiological relevance of this axis was confirmed by attenuated inflammation following either TET inhibition (dimethyloxallyl glycine, DMOG) or AIM2 deletion. These findings establish α-KG as a critical immunometabolic checkpoint in sepsis that licenses inflammatory cell death via TET2-mediated epigenetic control of AIM2. Our work not only elucidates a novel α-KG/TET2/AIM2 signaling axis in sepsis pathogenesis but also highlights the therapeutic potential of targeting this pathway to modulate immune responses.

Graphical Abstract

In sepsis, LPS challenge upregulates IDH1 expression, leading to the accumulation of α-KG. This increase in α-KG enhances dioxygenase activity of TET2, and promotes its recruitment to the AIM2 promoter, where it catalyzes local DNA demethylation and drives AIM2 transcription. The upregulated AIM2 protein then recruits the adaptor ASC to assemble the PANoptosome, a multiprotein complex that integrates key effector molecules from apoptosis, necroptosis, and pyroptosis pathways. This platform coordinately activates all three forms of programmed cell death, resulting in PANoptosis, a potent inflammatory cell death modality.Ultimately, this signaling cascade exacerbates multi-organ injury, a hallmark of severe sepsis.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12964-026-02740-3.

Introduction

Sepsis is a severe and life-threatening syndrome characterized by a dysregulated host response to infection, ultimately leading to excessive cytokine production and multiple organ dysfunction [1]. The innate immune system serves as the primary defense mechanism against pathogenic invasion. Macrophages, as a critical component of the innate immune response, are pivotal for antigen recognition, phagocytosis, and tissue homeostasis [2]. Recent studies indicate that macrophage death, particularly inflammatory forms involving membrane rupture (e.g., pyroptosis or necroptosis), releases substantial quantities of inflammatory mediators and cellular debris into the circulation. This process not only exacerbates cell death but also induces microcirculatory dysfunction, thereby contributing to multi-organ failure [3–7]. Physiologically regulated immune cell death is essential for maintaining microenvironmental equilibrium, whereas excessive death induces systemic immune dysregulation [8].

Multiple mechanisms may be involved in mediating the depletion of immune cell death in sepsis. Classical programmed cell death includes apoptosis, pyroptosis and necroptosis, which are historically described as independent signaling pathways. However, emerging evidence suggests extensive crosstalk among these pathways [6, 9, 10]. For example, cytokine storms in COVID-19 patients trigger concurrent activation of apoptosis, pyroptosis, and necroptosis in macrophages—a phenomenon coined PANoptosis [11, 12]. Similarly, herpes simplex virus-1 or Franciesia infection release pathogen- or damage-associated molecular patterns (PAMPs/DAMPs), activating inflammasome complexes to drive macrophage PANoptosis [13]. In sepsis, extensive macrophage death contributes to multi-organ dysfunction, though the potential role of PANoptosis in this context remains unexplored.

Physiological metabolite signaling is essential for maintaining immune homeostasis and dictating cell fate. Once metabolite signals are disturbed, the function and fate direction of immune cell are significantly affected [14]. It is well-established that macrophage metabolism transitions from oxidative phosphorylation to aerobic glycolysis during pathogenic challenge [15, 16]. Dysregulated glucose metabolism in macrophages results in the accumulation of specific metabolites, such as lactate, succinate, and fumarate, that directly modulate immune cell activity and survival [17–19]. For instance, lactate accumulation in the tumor microenvironment suppresses T cell function and promotes immune evasion [20], while succinate stabilizes HIF-1α to amplify pro-inflammatory responses in macrophages [21]. Research also demonstrates that fumarate activates innate immune responses via SNX9-dependent, MDV-mediated mtDNA release [22]. Although metabolite signaling is central to cellular communication, its role varies across pathological contexts. Thus, the identification of sepsis-specific metabolites regulating immune cell death remains crucial for developing targeted therapies. α-KG, as a key TCA cycle intermediate, functions beyond energy metabolism by modulating processes including epigenetic regulation, autophagy, cell death, and macrophage polarization [23–27]. Notably, recent studies reveal that α-KG can activate pyroptosis and necroptosis by promoting caspase-8-mediated cleavage of Gasdermin C (GSDMC) and modulating Receptor-interacting protein kinase 3 (RIPK-3) promoter methylation via TET enzymes [28, 29]. Despite its known involvement in regulating cell death pathways and macrophage function, the specific role of α-KG in driving macrophage death during sepsis remains unexplored.

This study aims to elucidate the molecular mechanisms and signaling pathways through which α-KG regulates macrophage PANoptosis in sepsis. Our findings reveal that α-KG accumulation in septic macrophages induces AIM2-PANoptosome assembly via TET2-dependent hypomethylation of the AIM2 promoter. Ultimately, this signaling cascade drives inflammatory macrophage death via PANoptosis, contributing to multi-organ failure and adverse clinical outcomes in sepsis. These findings shed light on the mechanisms underlying macrophage death in sepsis and provide a theoretical foundation and potential therapeutic targets for sepsis treatment.

Materials and methods

Mice and sepsis model

Wild-type C57BL/6 mice (6–8 weeks old, 20–25 g) used in this experiment were obtained from Shanghai SLAC Laboratory Animal Co., LTD. AIM2^−/−, Z-DNA-binding protein 1 deficient mice (ZBP1^−/−), Mediterranean fever ever deficient mice (MEFV^−/−), NACHT, LRR and PYD domains-containing protein 3 deficient mice (NLRP3^−/−), and IDH1^−/− mice were purchased from Shanghai Cyagen Biotechnology Co., LTD. All the mice were housed under specific pathogen-free conditions at the Animal Laboratory Center of the Shanghai General Hospital. The housing environment was maintained at 23–25℃ and 50%−65% humidity with 50–65% relative humidity under a 12-h light/dark cycle (lights on at 07:00, off at 19:00). All experimental protocols were approved by the Ethics Committee of Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine (Ethical approval number: 2023AW055) and conducted strictly in compliance with institutional animal welfare guidelines.

Endotoxemia model

Mice (6–8 weeks old, 20–25 g) neally injected with LPS (15 mg/kg; Sigma-Aldrich, #L2630). Tissues were collected at designated time points post-injection for analysis.

Cecal ligation and puncture (CLP) model

Mice were anesthetized with sevoflurane and immobilized in a supine position. After disinfecting the abdomen using 75% ethanol, a midline laparotomy (1–2 cm) was performed to access the cecum. The cecum was found and ligated with 4–0 silk. Then, pierce the cecum with a 25-gauge syringe needle. After gently extruding a small volume of fecal content, the cecum was returned to the abdominal cavity, and the peritoneum and skin were sutured in layers. Postoperatively, mice received subcutaneous injection of 1 mL sterile saline and were placed on a warming pad for recovery. Tissues were harvested at predetermined time points for further analysis.

Clinical severity scores

The clinical score of each animal was evaluated based on a previous study. Briefly, the following scoring criteria (in points). [a]general appearance: normal (0 point), lack of grooming (1 point), piloerection (2 points), hunched posture (3 points), above and half-closed eyes (4 points); [b] behavior—unprovoked: normal (0 point), minor changes (1 point), reduced mobility and isolation (2 points), restless or unusually still (3 points); behaviour—provoked: responsive and alert (0 point), unresponsive and not alert (3 points); [c] clinical signs: normal respiratory rate (0 point), slight changes (1 point), decreased rate with abdominal breathing (2 points), pronounced abdominal breathing and cyanosis (3 points); [d] hydration status: normal (0 point), dehydrated (3 points). In murine sepsis models, a higher clinical score correlates strongly with greater disease severity and poorer outcomes.

Cell culture and treatments

Peripheral mononuclear macrophages (PBMCs) were separated from blood samples by using human peripheral blood mononuclear cell isolation medium (P9010, Solarbio) or mouse peripheral blood mononuclear cell isolation medium (P6340, Solarbio). Primary mouse bone marrow-derived macrophages (BMDMs) from the bone marrow of wild-type (WT) and indicated knockout mice were grown for 6 days in RPMI1640 containing macrophage colony-stimulating factor (10 ng/ml; #100–21-10, PeproTech), 10% fetal bovine serum (FBS), and 1% penicillin–streptomycin antibiotics (NCM Biotech, C100C5). Peritoneal exudate macrophage (PEM) was isolated from the peritoneal lavage fluid of mice intraperitoneally injected with 4% thioglycolate broth and cultivated in dulbecco's modified eagle's medium (DMEM) supplemented with 10% FBS and 1% penicillin–streptomycin antibiotics. Cells were seeded into 12 well plates at a concentration of 1 × 10⁶ cells/well and incubated with growth media overnight before use. RAW 264.7 cells were obtained from the cell bank of the Chinese Academy of Sciences and cultured in DMEM with 10% FBS and 1% penicillin–streptomycin antibiotics. Cells were stimulated with 500 ng/ml LPS, 5 mM DM-α-KG, 50 μM mM DMOG, or 10 μM IDH1-305 at the indicated time.

RNA interference and plasmid transfection

The mouse siRNA against TET1, TET2 and TET3 genes and negative control siRNA were purchased from Genechem. Sequences of siRNAs were listed as follows: TET1 siRNA 5′- CUCCUAUCAAUCAGAUAAATT-3′; TET2 siRNA 5′- GGCUGUCAAACUCCAGAAUTT-3′; TET3 siRNA 5′- CCUGAAGUCAGAGGAGAAATT-3′. According to the manufacturer’s protocols, BMDMs were transfected with siRNA (20 nM) using Lipofectamine 2000 (Invitrogen). The efficiency of siRNA was validated by RTqPCR examination.

Lentiviral-mediated AIM2 over-expression

The lentiviral constructs (PGMLV-CMV-Mouse_Aim2-EF1-ZsGreen1-T2A-Puro and the empty vector PGMLV-CMV-EF1-ZsGreen1-T2A-Puro) were generated by Genomeditech (Shanghai). To determine the optimal transduction conditions, AIM2^−/− BMDMs were transduced with a range of multiplicities of infection (MOI from 10 to 100). Based on transduction efficiency and cell viability, an MOI of 50 was selected for all subsequent experiments. AIM2 expressions were confirmed by fluorescence microscopy and Western blot analysis.

Real-time PCR analysis

Total RNA was extracted from cells or tissues at the indicated time points by using TRIzol reagent (Takara, T9108). 100 ng RNA of each sample was reverse transcribed into cDNA using SuperScript®III Kit (Thermo Scientific, 18,080,093) according to the manufacturer’s protocols. Then cDNA was applied to qPCR experiments. Comparative analysis of relative expression based on the ΔΔCT method was used to analyze the results. The PCR primers used in this study were provided as followed (5′−3′): 18S: Forward, 5’- TTCCGATAACGAACGAGACTCT—3’; Reverse, 5’- TGGCTGAACGCCACTTGTC—3’. IL-1β: Forward, 5’- GAAATGCCACCTTTTGACAGTG—3’; Reverse, 5’- TGGATGCTCTCATCAGGACAG—3’. IL-6: Forward, 5’- CTGCAAGAGACTTCCATCCAG—3’; Reverse, 5’- AGTGGTATAGACAGGTCTGTTGG—3’. IL-18: Forward, 5’- GACTCTTGCGTCAACTTCAAGG- 3’; Reverse, 5’- CAGGCTGTCTTTTGTCAACGA- 3’. TNFα: Forward, 5’-CCTGTAGCCCACGTCGTAG-3’; Reverse, 5’-GGGAGTAGACAAGGTACAACCC-3’. AIM2: Forward, 5’-GTCACCAGTTCCTCAGTTGTG-3’; Reverse, 5’-CACCTCCATTGTCCCTGTTTTAT-3’. ZBP1: Forward, 5’-AAGAGTCCCCTGCGATTATTTG-3’; Reverse, 5’-TCTGGATGGCGTTTGAATTGG-3’. PYRIN: Forward, 5’-TCATCTGCTAAACACCCTGGA-3’; Reverse, 5’-GGGATCTTAGAGTGGCCCTTC-3’. NLRP3: Forward, 5’-ATTACCCGCCCGAGAAAGG −3’; Reverse, 5’-TCGCAGCAAAGATCCACACAG-3’. TET1: Forward, 5’-ACACAGTGGTGCTAATGCAG-3’; Reverse, 5’-AGCATGAACGGGAGAATCGG −3’. TET2: Forward, 5’-AGAGAAGACAATCGAGAAGTCGG-3’; Reverse, 5’-CCTTCCGTACTCCCAAACTCAT- 3’. TET3: Forward, 5’- TGCGATTGTGTCGAACAAATAGT-3’; Reverse, 5’-TCCATACCGATCCTCCATGAG-3’.

Immunoblot analysis

The tissues and cells were lysed with cell lysate (containing 1 × protease inhibitor and 1 × phosphatase inhibitor). After centrifugation, the supernatant was separated and added with 100 μl 5 × SDS sample loading buffer. Each sample hole was loaded with 30 μg of protein sample. Proteins of different molecular weights were dissolved and separated by 8%−12% polyacrylamide gel electrophoresis. After the protein was electrophoretically transferred to a PVDF membrane (Millipore, IPVH00010). Non-specific binding was blocked by incubation with 5% skim milk at 4 ℃ for 2 h; then membranes were incubated with the following primary antibodies at 4 ℃ overnight: Caspase1 (1:2000,Cell Signaling Technology, #24,232), Cleaved-Caspase1 (1:1000,Cell Signaling Technology, #89,332), Gasdermin D (GSDMD) (1:2000,Cell Signaling Technology, #39,754), Cleaved-GSDMD (1:1000,Cell Signaling Technology, #10,137), Caspase3 (1:2000,Cell Signaling Technology, #14,220), Cleaved-Caspase3 (1:1000,Cell Signaling Technology, #9661), Caspase7 (1:2000,Cell Signaling Technology, #9497), Cleaved-Caspase7 (1:1000,Cell Signaling Technology, #8438), receptor-interacting protein kinase 1 (RIPK-1) (1:2000,Cell Signaling Technology, #3493), phosphorylated RIPK1 (pRIPK-1) (1:1000,Cell Signaling Technology, #31,122), mixed lineage kinase domain-like pseudokinase (MLKL) (1:2000,Cell Signaling Technology, #37,705), phosphorylated MLKL (p-MLKL) (1:1000, abcam, ab196436), AIM2 (1:2000, Cell Signaling Technology, #63,660), Pyrin (1:2000, abcam, ab195975), TET1(1:1000, GeneTex, GTX124207), TET2 (1:1000, abcam, ab124297), TET3 (1:1000, GeneTex, GTX121453-S), apoptosis-associated speck-like protein containing a CARD (ASC) (1:1000, Santa Cruz, sc-514414), ASC (1:1000, abcam, ab175449), ZBP1 (1:1000, Novus biologicals, NBP2-80,056), NLRP3 (1:2000, Cell Signaling Technology, #15,101), ZBP1 (1:1000, Novus biologicals, NBP2-80,056), β-tubulin(1:5000, ABclonal, A12289). After washing, the membranes were incubated with the corresponding horseradish peroxidase at room temperature for 2 h.

Immunoprecipitation (IP)

After washing the cells with cold phosphate-buffered saline (PBS), BMDMs was lysed on ice with an IP lysate buffer (containing protease inhibitors and phosphatase inhibitors) for 30 min. After centrifugation at 12,000 rpm for 15 min at 4 ℃, the supernatant was collected. For endogenous IP, the supernatant was added with Protein-A/G agarose beads and designated antibodies, and then incubated at 4 °C overnight. The agarose beads were harvested the next day and washed 5 times with IP buffer. Immunoprecipitated proteins were boiled in 1 × SDS loading buffer at 100 °C for 5 min. Then western blotting was performed to evaluate all samples.

Immunofluorescence staining

After washing the cells with cold PBS for 3 times, the cells were fixed in 4% paraformaldehyde at room temperature for 20 min. Then, the fixed cells were permeabilized with 0.25% Triton X-100 for 15 min. Next, the cells were blocked with 1% bovine albumin (BSA) for 2 h and incubated by the indicated primary antibody as followed overnight at 4℃: ASC (1:500, abcam, ab175449), Cleaved-Caspase3 (1:500, Cell Signaling Technology, #9661), p-MLKL (1:800, abcam, ab196436), AIM2 (1:1000, Cell Signaling Technology, #63,660). After washing with PBST (0.05% Tween 20 in PBS) for three times, cells were incubated with the following secondary antibody at room temperature for 2 h and stained with DAPI to show the nucleus. Images were obtained using a laser scanning confocal microscope (Lecia, SP8).

Hematoxylin and eosin (H&E) staining

The lungs, liver, and kidney tissues were washed with PBS and soaked in 10% formalin at room temperature for 24 h. The tissues were then embedded in paraffin wax and cut into 5 μm sections. Sections were stained with H&E. Images were taken by a microscope (Lecia, DMi8).

Lung injury scores

The evaluation was conducted by two blinded assessors using the Smith Lung Injury Score system, which evaluates four parameters: alveolar and interstitial hemorrhage, alveolar and interstitial inflammation, pulmonary edema, and pulmonary atelectasis, along with hyaline membrane formation. Scores were calculated based on all components observed in randomized fields. The grading criteria are defined as: 0 points for no pathological changes in lung tissue; 1 point for lesions affecting < 25% of lung tissue; 2 points for 25–50% involvement; 3 points for 50–75% involvement; and 4 points for > 75% involvement. The total lung injury score represents the sum of all four component scores.

LDH release assay

According to the manufacturer’s protocols for Lactate dehydrogenase assay kit (Nanjing jiancheng, A020-2–1), the lactate dehydrogenase (LDH) in the cellular supernatant was detected to indicate dead cells.

α-ketoglutarate Quantitation assay

According to the manufacturer’s instructions (Sigma-Aldrich, MAK054), the quantitation of α-ketoglutarate in cells (2 × 10⁶) was detected. α-KG concentration is determined by a coupled enzyme assay, which results in a colorimetric (570 nm) product. The α-KG concentrations in PBMCs of clinical participants were detected from healthy controls and septic patients. This study involving human participants was reviewed and approved by the Ethics Committee of Shanghai General Hospital School of Medicine (No.2024SQ313). All procedures were performed in accordance with the ethical standards of the Declaration of Helsinki. Informed consent was obtained from all participants or their legal guardians. Relevant sample information is provided in the supplementary materials (Supplementary Table 1) to ensure transparency.

ELISA

The release of IL-1β, IL-6, and TNF-α in the supernatant was analyzed using an enzyme-linked immunosorbent assay (ELISA). ELISA kits were purchased from Invitrogen. According to the manufacturer's instructions, equal amounts of suitably diluted cell supernatant were added to each plate and absorbance at 450 nm was measured. The corrected absorbance value was obtained by subtracting the background absorbance, and then the cytokine concentration was calculated by extrapolation from the standard curves drawn using GraphPad Prism 9.2.0.

Chromatin immunoprecipitation assays (ChIP)

According to the manufacturer’s instruction (Sigma-Aldrich,17–371), genomic DNA and nuclear protein were crosslinked by adding 10 ml 1% formaldehyde and incubating at room temperature for 10 min. 1 ml 10 × glycine was added to terminate crosslinking and incubated at room temperature for 10 min. Cells were collected, centrifuged, and re-suspended in a lysate buffer (1 ml SDS lysate, 5 μl protease inhibitor mixture Ⅱ). The cell lysate was treated with ultrasound and centrifugated with immunoprecipitation Buffer (900 μl ChIP Dilution Buffer, 4.5 μl protease inhibitor mixture II). The chromatin resuspension was incubated with protein G agarose beads at 4 °C for 1 h, and the recovered chromatin solution was incubated with 1 μg of the specified antibody at 4 °C overnight. Add 60 μl protein A/G agarose beads and rotate at 4 °C for 2 h. 1 μg of the specified antibody was added to the recovered chromatin solution and rotated overnight at 4 °C. On the second day, 60 μl agarose beads were added to each group of chromatin solution and rotated at 4 ℃ for 1 h. After centrifugation, agarose beads were washed with the following buffers successively: low salt buffer, high salt buffer, LiCl wash buffer and TE buffer. The protein/DNA complex was eluted by adding 200 μl elution buffer (ddH₂O 170 μl, 1 M NaHCO₃ 20 μl, 20%SDS 10 μl). 5 M NaCL 8 μl was added and incubated at 65 ℃ overnight. 0.5 M EDTA 4 μl, Tris–HCL 8 μl and protease K 1 μl were added and incubated at 45 ℃ for 1 h. The DNA was purified and extracted. Genomic DNA was amplified by PCR using specific primers. AIM2 primer: Forward, 5’- TCAGCAAACCTAAGGGAGAG-3’; Reverse, 5’-GGAGTGATGACATCACCACT-3’. PCR product was isolated by 2% agarose gel electrophoresis.

Statistical analysis

Data was input into the GraphPad Prism software for statistical analysis. The two groups of data were analyzed using t test method. Two-way ANOVA and Tukey's multiple comparisons test were used for the analysis of multiple comparisons. The survival rate was analyzed by Kaplan–Meier analysis. All experimental results are displayed as Mean ± SD (i.e. mean ± standard deviation).

Results

Sepsis induces α-KG accumulation and IDH1 upregulation

To systematically investigate metabolic perturbations in sepsis pathogenesis, an unbiased liquid chromatography-mass spectrometry (LC–MS)-based metabolomic analysis was conducted on inflammatory PBMCs derived from a murine CLP model. Among the most significantly upregulated metabolites in the TCA cycle, α-KG was identified as one of the prominent metabolites induced upon stimulation, along with previously established metabolites such as itaconate (Fig. 1a-f). Clinical evidence further supported this phenomenon, showing significantly higher α-KG concentrations in PBMCs from septic patients than in those from healthy controls. (Fig. 1g). In septic mice model, we also observed that endotoxemia or CLP promoted the production of α-KG in a time-dependent manner (Fig. 1h-i).

Fig. 1.

Sepsis induced upregulation of IDH1-α-KG. a Schematic diagram of the TCA cycle; b Metabolomic screening of metabolite level changes in mouse PBMCs (n = 4); c α-KG levels in PBMCs (n = 4); d Citric acid levels in PBMCs (n = 4); e Succinic acid levels in PBMCs (n = 4); f Itaconic acid levels in PBMCs (n = 4); g α-KG levels in PBMCs of sepsis (n = 16) and healthy (n = 11) patient; h Changes in PBMCs α-KG levels following LPS intraperitoneal injection (n = 5); i. Changes in PBMCs α-KG levels in CLP-induced septic mice (n = 5); j, k IDH1 expression in the GSE54514 dataset; l IDH1 protein expression in lung tissues of endotoxemia and CLP models (n = 3); m, n IDH1 mRNA expression in lung tissues of endotoxemia and CLP models (n = 6). o Immunohistochemical detection of IDH1 expression in mouse lung tissues

Given the enzymatic regulation of α-KG biosynthesis by isocitrate dehydrogenase (IDH) [30, 31], we subsequently analyzed IDH1 expression using a transcriptomic dataset (GEO accession: GSE54514) encompassing sepsis survivors and non-survivors. Bioinformatic analysis revealed significant IDH1 upregulation in fatal sepsis cases (Fig. 1j-k). Experimental validation confirmed concordant elevation of IDH1 expression at both transcriptional and translational levels in PBMCs of endotoxemia and CLP mice, respectively (Fig. 1l-n). Similarly, the positive staining of IDH1 in the lung tissues of CLP mice was significantly higher than that in WT lung tissues (Fig. 1o). Collectively, these findings demonstrate elevated IDH1 expression with concomitant α-KG accumulation in the pathogenesis of sepsis.

α-KG exacerbates sepsis-associated multiorgan dysfunction

The pathophysiological significance of α-KG accumulation in sepsis prognosis and its mechanistic underpinnings remain incompletely elucidated. To systematically investigate this phenomenon, murine models were subjected to LPS challenge with or without DM-α-KG coadministration. DM-α-KG administration exacerbated clinical severity scores and mortality rates in LPS-challenged mice (Fig. 2a-b), as well as promoted multi-organ injury, including lung, liver, and kidney (Fig. 2c-d). Consistently, levels of hepatic injury markers aspartate aminotransferase (AST) and alanine aminotransferase (ALT), and renal dysfunction indicators blood urea nitrogen (BUN) and creatinine (CRE) were significantly increased in the co-treatment group (Fig. 2e-h). Meanwhile, systemic pro-inflammatory cytokine storm was observed, with serum concentrations of IL-6, IL-1β, and TNFα exceeding LPS group levels (Fig. 2i). Importantly, genetic ablation of IDH1 markedly attenuated the clinical scores and mortality of endotoxemia mice, as well as multi-organ injury (Fig. 2j-q, Fig. S2d). Mechanistically, IDH1 deficiency abrogated α-KG-mediated cytokine amplification, reducing IL-6, IL-1β, and TNF-α levels (Fig. 2r). These findings collectively establish α-KG accumulation as a pathological determinant of sepsis progression, while highlighting IDH1-mediated metabolic reprogramming as a critical yet underexplored regulatory axis requiring elucidation through mechanistic investigations.

Fig. 2.

α-KG exacerbates multiorgan dysfunction in sepsis a Clinical state of mice following LPS and DM-α-KG intraperitoneal injection (n = 10); b Survival rate of mice intraperitoneally injected with LPS and DM-α-KG (n = 10); c Lung injury score of mice injected with LPS and DM-α-KG (n = 5); d H&E staining in lung tissues, liver tissues, and kidney tissues of mice; e–h AST, ALT, BUN and CRE levels in serum of mice (n = 4); i IL-6, IL-1β and TNFα levels in serum of mice (n = 3); j Clinical state of WT and IDH1^−/− mice k Survival rate of WT and IDH1^−/− mice intraperitoneally injected with LPS (n = 5 for PBS, n = 10 for LPS); l Lung injury score of mice WT and IDH1^−/− mice intraperitoneally injected with LPS (n = 5); m H&E staining in lung tissues, liver tissues, and kidney tissues of WT and IDH1^−/− mice (n = 5); n-q AST, ALT, BUN and CRE levels in serum of WT and IDH1^−/− mice (n = 3); r IL-6, IL-1β and TNFα levels in serum of WT and IDH1.^−/− mice (n = 3)

α-KG mediates inflammatory cell death

Excessive activation of innate immunity by PAMPs and DAMPs drives inflammatory cell death, culminating in tissue damage and multiorgan failure. α-KG as a metabolism-associated molecular patterns (MAMPs, subclass of DAMPs [32]) may mediate sepsis-associated organ injury and mortality through inflammatory cell death mechanisms. In vitro, we co-treated BMDMs, PEM and RAW264.7 cell line with LPS and DM-α-KG, interestingly, significantly high levels of cell death were observed in BMDMs, PEM and RAW264.7 cell line by Sytox staining (Fig. 3a-b, S1a), as well as LDH release in BMDMs (Fig. S1b).

Fig. 3.

DM-α-KG and LPS induces cell death and pro-inflammatory cytokines. a Sytox staining of dead cell in bone marrow derived macrophages (BMDMs, n = 5); b Sytox staining of dead cell in peritoneal macrophage (PEM, n = 3); c IL-1β, IL-6 and TNFα levels in cellular supernatant of BMDMs (n = 5); d IL-1β, IL-6 and TNFα mRNA levels in BMDMs (n = 4); e IL-1β, IL-6 and TNFα levels in cellular supernatant of BMDMs treated with LPS, DM-α-KG and IDH1-305 (n = 3); f IL-1β, IL-6 and TNFα mRNA levels in BMDMs treated with LPS, DM-α-KG and IDH1-305 (n = 6)

The process of cell death is associated with the release of a large amounts of proinflammatory cytokines. Co-treated BMDMs with LPS plus DM-α-KG significantly elevated IL-1β, IL-6, and TNF-α secretion (Fig. 3c) and corresponding mRNA levels (Fig. 3d, S1c). While blocking the production of α-KG by inhibiting IDH1 with IDH1-305 markedly attenuated the release of IL-1β, IL-6 and TNF-α and transcriptional activation (Fig. 3e-f), confirming α-KG's role in potentiating inflammatory responses. These results collectively demonstrate that α-KG synergizes with PAMPs to drive macrophage inflammatory death and cytokine storm during sepsis.

LPS and DM-α-KG mediates PANoptosis of macrophage depending on AIM2

The molecular mechanisms underlying DM-α-KG-induced inflammatory cell death remain to be fully characterized. Pattern recognition receptors (PRRs), which orchestrate transcriptional activation of inflammatory cytokines through detection of PAMPs and DAMPs, critically regulate innate immunity, adaptive responses, and programmed cell death (PCD) modalities including pyroptosis, necroptosis, and apoptosis [33]. To delineate the cell death pathways activated by DM-α-KG, we elucidate the biochemical characteristics of cell death induced in BMDMs treated with LPS and different concentrations of DM-α-KG. Caspase-1 cleavage was upregulated in response to LPS and high concentration of DM-α-KG treatment (Fig. S2a), indicative of inflammasome activation. Downstream of inflammasome activation, the N-terminal of gasdermin D (GSDMD) can be processed to form P30 fragment, which is released and executes cell death. Cleaved-GSDMD were increased in LPS-stimulated BMDMs treated with 5 mM and 10 mM DM-α-KG (Fig. S2a). Additionally, cleavage of Caspase-7 and Caspase-3 was induced to a greater extent by treatment with 5 mM DM-α-KG than by a lower dose. Furthermore, the phosphorylation of necroptotic molecules RIPK-1 and MLKL was significantly increased in response to LPS and DM-α-KG (Fig. S2a).

Next, to examine the synergistic contribution of LPS and DM-α-KG in triggering these cell death pathways, we administrated BMDMs with LPS/DM-α-KG co-treatment. As the immunoblotting results shown, the combination of LPS and DM-α-KG induced robust GSDMD cleavage, as well as caspase1, caspase3, and caspase7 activation. Nevertheless, the co-treatment of LPS and DM-α-KG induced robust phosphorylation of RIPK-1 and MLKL, but not LPS or DM-α-KG alone (Fig. 4a). Consistently, we observed similar results in lung tissue from mice that had been intraperitoneally injected with LPS and DM-α-KG (Fig. S2b). Meanwhile, we monitored cellular morphology of BMDMs treated by LPS and DM-α-KG under the microscope. Various morphologies of dead macrophage were observed, including cell swelling and bulbous protrusion (1), cell membrane blistering (2), cell contraction and cell collapse (3), cell lysis and overflow of cell contents (4) (Fig. S2c). Taken together, these in vivo and in vitro results suggested that synergistic signaling by LPS plus DM-α-KG treatment induces a cell death signature consistent with PANoptosis, rather than individually activating pyroptosis, necroptosis, or apoptosis.

Fig. 4.

DM-α-KG and LPS co-treatment induce AIM2-dependent PANoptosis. a Western blot of Caspase-1, GSDMD, Caspase3, Cleaved-caspase3, Caspase7, Cleaved-caspase7, MLKL, pMLKL, pRIPK-1, total RIPK-1(tRIPK-1) and β-tubulin in BMDMs; b Immunofluorescence images of ASC⁺pMLKL⁺active-Caspase3⁺ specks in BMDMs (n = 3);c IP in BMDMs followed by immunoblot analysis of Pyrin, NLRP3, ZBP1, ASC, AIM2 and β-tubulin expression; d Sytox staining of AIM2^−/−, Pyrin^−/−, ZBP1^−/−, NLRP3^−/− and WT BMDMs treated with LPS and DM-α-KG (n = 5); e Immunofluorescence of ASC⁺pMLKL⁺active-Caspase3⁺ specks in AIM2^−/−, Pyrin^−/−, ZBP1^−/−, NLRP3^−/− and WT BMDMs (n = 3); f Immunoblot analysis of GSDMD, Caspase3, MLKL, and pMLKL in the lung tissue of WT and AIM2^−/− mice

Mechanistically, PRR-mediated PANoptosis requires assembly of a supramolecular PANoptosome complex, wherein sensor proteins (e.g., AIM2, ZBP1, Pyrin) engage adaptor ASC via homotypic pyrin domain (PYD) or caspase activation and recruitment domain (CARD) interactions, facilitating caspase recruitment [33]. Immunofluorescence microscopy confirmed ASC speck formation with spatial colocalization of cleaved caspase-3 and phosphorylated MLKL in LPS/DM-α-KG-stimulated BMDMs (Fig. 4b). Co-immunoprecipitation assays further validated PANoptosome assembly, demonstrating physical interactions among ASC, Pyrin, NLRP3, ZBP1, and AIM2 (Fig. 4c).

To identify critical upstream sensors governing this pathway, genetic ablation studies were performed. We evaluated cell death in AIM2-, Pyrin-, ZBP1- and NLRP3-deficient BMDMs stimulated with LPS plus DM-α-KG (Fig. S2d). While NLRP3 deficiency failed to attenuate cell death, partial protection was observed in ZBP1^−/− and Pyrin^−/− BMDMs. Strikingly, AIM2 knockout conferred robust cytoprotection (Fig. 4d). Consistently, AIM2 ablation disrupted ASC-caspase-3-MLKL colocalization (Fig. 4e) and abrogated GSDMD/caspase-3/MLKL activation in lung tissues of mice in response to LPS and DM-α-KG stimulation (Fig. 4f). To definitively establish the requirement of AIM2 for LPS/DM-α-KG-induced PANoptosis, we performed rescue experiments through lentiviral-mediated AIM2 overexpression (Fig.S3a). AIM2 was overexpressed in AIM2^−/−BMDMs. Western blot analysis demonstrated that restoration of AIM2 expression in AIM2^−/− BMDMs substantially rescued the activation of key PANoptosis executors, including cleaved GSDMD, cleaved caspase-3, and phosphorylated MLKL, which were otherwise suppressed in AIM2-deficient cells (Fig.S3b). Consistent with this, the release of inflammatory cytokines including IL-1β, IL-6, and TNF-α was significantly enhanced upon AIM2 reconstitution (Fig.S3c). Collectively, these findings establish AIM2-dependent PANoptosome assembly as the central regulatory mechanism connecting α-KG accumulation to inflammatory cell death in sepsis pathogenesis.

IDH1 suppresses LPS- and DM-α-KG-induced AIM2 expression

Quantitative analyses using immunoblotting and qPCR revealed differential regulation of cytosolic sensor expression under inflammatory stimuli. Transcriptional upregulation of Pyrin, NLRP3, and ZBP1 was selectively induced by LPS challenge, whereas DM-α-KG group failed to elicit significant transcriptional activation. Quantitative immunoblot analysis demonstrated comparable protein expression of Pyrin, NLRP3, and ZBP1 between LPS/DM-α-KG co-treatment and LPS group (Fig. 5a-e). Notably, both LPS and DM-α-KG independently induced a robust elevation of AIM2 expression at both the mRNA and protein levels, with synergistic enhancement observed under co-treatment conditions (Fig. 5a-e). Pharmacological inhibition of IDH1 abolished LPS-driven AIM2 transcriptional activation (Fig. 5f), while genetic IDH1 depletion significantly attenuated AIM2 expression in pulmonary tissues post-LPS challenge (Fig. 5g). Immunofluorescence confirmed increased AIM2-ASC co-localization speck formation in response to LPS or DM-α-KG treatment, which was amplified synergistically by co-treament of LPS and DM-α-KG (Fig. S4a). In addition, compared with WT BMDMs, we observed reduced specks of AIM2⁺ASC⁺ in IDH^−/− BMDMs in response to LPS (Fig. S4b). Collectively, these findings establish α-KG as a metabolic regulator of AIM2 inflammasome activation in macrophages through an IDH1-dependent transcriptional mechanism.

Fig. 5.

α-KG primes AIM2 via TET2-Mediated DNA demethylation. a Immunoblot of AIM2, Pyrin, ZBP1, and NLRP3 protein levels in BMDMs administrated with LPS and DM-α-KG; b-e AIM2, Pyrin, ZBP1, and NLRP3 mRNA level in BMDMs administrated with LPS and DM-α-KG (n = 6); f AIM2 mRNA level in BMDMs administrated with LPS and IDH1-305 (n = 6); g AIM2 mRNA level in WT and IDH1^−/− BMDMs administrated PBS or LPS (n = 6); h Pyrosequencing of AIM2 methylation level in lung tissue of mice treated with LPS and DM-α-KG (n = 10); i Pyrosequencing of AIM2 methylation level in lung tissue of CLP mice (n = 10); j TETs enzyme activity in BMDMs pretreated with IDH1-305 followed by administration of LPS and DM-α-KG (n = 5); k Immunoblot of AIM2 in WT BMDMs pretreated with DMOG followed by administration of LPS and DM-α-KG; l AIM2 mRNA level in BMDMs administrated with LPS, DM-α-KG, and DMOG (n = 6); m Sytox staining of BMDMs treated with DMOG, LPS, and DM-α-KG (n = 5); n Sytox staining of BMDMs transfected with TET1siRNA, TET2 siRNA or TET3 siRNA, following the treatment of LPS and DM-α-KG (n = 5); o Western blot of AIM2 in TET2 siRNA-transfected BMDMs with treatment of LPS and DM-α-KG; p AIM2 mRNA level in TET2 siRNA-transfected BMDMs with treatment of LPS and DM-α-KG (n = 6); q ChIP analysis of BMDMs administrated with LPS, DM-α-KG or IDH1-305

TET2 drives PANoptosis induced by DM-α-KG and LPS via DNA demethylation of AIM2

The above data illustrated a unique α-KG-dependent signal transduction pathway proceeding from the elevation of AIM2 expression to PANoptosis induction. While α-KG is classically recognized as a central metabolic intermediate, emerging evidence implicates its non-canonical role as a co-substrate for α-KG-dependent dioxygenases, including TET-family DNA demethylases, Jmjc-domain histone modifiers, and prolyl hydroxylases [34]. Notably, TET enzymes have been mechanistically associated with inflammasome activation through epigenetic remodeling of sensor gene promoters [35]. To elucidate α-KG-mediated AIM2 transcriptional regulation, pyrosequencing analysis of lung tissues revealed hypomethylation at the AIM2 promoter in LPS/DM-α-KG-co-treated and CLP murine models (Fig. 5h-i). Functional studies demonstrated that LPS priming enhanced TET enzymatic activity in macrophages, which was further potentiated by DM-α-KG co-stimulation. Pharmacological inhibition of α-KG biosynthesis via IDH1-305 attenuated this activation, confirming IDH1-dependent metabolic-epigenetic crosstalk (Fig. 5j). To explore whether the induction of AIM2 transcription in response to LPS and DM-α-KG depends on TET2 enzyme activity, we pretreated BMDMs with DMOG (a widely used inhibitor of TETs enzyme). Immunoblot and qPCR analysis showed that upregulated AIM2 expression induced by LPS and DM-α-KG was blocked by the pretreatment of DMOG in BMDMs (Fig. 5k-l). To understand the role of TETs in cell death induced by LPS and DM-α-KG, BMDMs were pretreated with DMOG before the stimulation of LPS plus DM-α-KG. We observed that BMDMs pretreated with DMOG were 80% protected from death during LPS and DM-α-KG treatment (Fig. 5m). The TETs demethylase family includes TET1, TET2, and TET3. To further screen which TETs enzymes mediated macrophage death, TET1siRNA, TET2siRNA, and TET3siRNA were constructed, respectively (Fig. S4c-d). BMDMs were transfected with wild-type siRNA, TET1-siRNA, TET2-siRNA, or TET3-siRNA, then treated with LPS and DM-α-KG. The results indicated that TET2siRNA obviously blocked cell death by LPS and DM-α-KG treatment (Fig. 5n). Moreover, reduction in TET2 mRNA levels by siRNA significantly suppressed AIM2 expression (Fig. 5o-p). In addition, ChIP assays verified TET2 recruitment to the AIM2 promoter, amplified by α-KG co-stimulation and abrogated by IDH1 inhibition (Fig. 5q).

We therefore explored whether upregulated TETs activity played a role in AIM2-dependent PANoptosome assembly. Immunofluorescence results revealed DMOG-mediated suppression of ASC-caspase3-MLKL colocalization (Fig.S4e). In addition, DMOG reduced the level of cleaved-GSDMD, cleaved-caspase3, and pMLKL in BMDMs stimulated by LPS and DM-α-KG (Fig. S4f), and attenuated IL-1β/IL-6/TNF-α secretion (Fig. S4g-h). Similarly, TET2 knockdown weakened fluorescence intensity and co-localization of ASC, cleaved caspase-3, and pMLKL (Fig. S4i). Immunoblotting analysis showed that the cleavage and activation of GSDMD, Caspase3 and MLKL were hindered by TET2 knockdown in BMDMs treated with LPS plus DM-α-KG (Fig. S4j). These findings collectively establish TET2-driven DNA demethylation as the epigenetic mechanism underlying α-KG/AIM2-mediated PANoptosis.

Blocking TET2-AIM2 axis reduces mortality in mice co-treated by LPS plus DM-α-KG

We then determined whether the observed disease pathogenesis in LPS plus DM-α-KG treated- mice model was driven by TET2-AIM2 axis. In mice injected with LPS plus DM-α-KG, we observed that DMOG pretreatment improved survival rates, reduced clinical severity scores, and attenuated multiorgan dysfunction (Fig. 6a-h). Meanwhile, DMOG inhibited the release of inflammatory cytokines induced by LPS and DM-α-KG (Fig. 6i). Further, AIM2-deficient mice demonstrated complete protection against LPS/DM-α-KG-induced lethality (Fig. 6j-k). Similarly, there was a significant decrease in the extent of damage in AIM2^−/− lung, liver, and kidney tissue (Fig. 6l-q). The inflammatory cytokine levels were reduced in serum of AIM2^−/− mice (Fig. 6r). These critical findings were further confirmed in CLP model. Specifically, our experiments showed that both genetic AIM2 knockout and pharmacological TET2 inhibition with DMOG significantly improved survival and attenuated organ injury in CLP mice (Fig.S4k-l). These findings conclusively demonstrate that TET2/AIM2 axis disruption confers robust protection against metabolic-inflammatory cell death and mortality in experimental sepsis.

Fig. 6.

Inhibiting TET2-AIM2 axis provides protection against injury drived by co-treatment of LPS and DM-α-KG. a Clinical state of mice following LPS,DM-α-KG and DMOG intraperitoneal injection (n = 5 for DMSO, n = 10 for other group); b Survival rate of mice intraperitoneally injected with LPS, DM-α-KG, and DMOG (n = 5 for DMSO, n = 10 for other group); c Lung injury score of mice injected with LPS, DM-α-KG, and DMOG (n = 5); d H&E staining in lung tissues, liver tissues, and kidney tissues of mice intraperitoneally injected with LPS, DM-α-KG, and DMOG (n = 3); e–h AST, ALT, BUN, and CRE levels in serum of mice (n = 3); i IL-1β, IL-6, and TNFα levels in serum of mice (n = 3); j Clinical state of WT and AIM2^−/− mice following LPS and DM-α-KG intraperitoneal injection; k Survival rate of WT and AIM2^−/− mice intraperitoneally injected with LPS and DM-α-KG (n = 5 for DMSO, n = 10 for LPS + DM-α-KG); l Lung injury score of WT and AIM2^−/− mice intraperitoneally injected with LPS and DM-α-KG (n = 5); m H&E staining in lung tissues, liver tissues, and kidney tissues of WT and AIM2.^−/− mice (n = 3); n-q. AST, ALT, BUN, and CRE levels in serum of mice (n = 3); r IL-1β, IL-6, and TNFα levels in serum of mice (n = 3)

Discussion

While accumulating evidence implicates metabolic dysregulation in sepsis pathophysiology [36, 37], the precise regulatory mechanisms remain incompletely characterized. The study uncovers a previously unrecognized mechanism through which metabolic reprogramming in sepsis drives lethal inflammatory cell death via an integrated IDH1/α-KG/TET2/AIM2 axis, providing new insights into the metabolic‐immunological crosstalk that drives disease progression. Our investigation systematically examines α-KG accumulation as a critical mediator of macrophage PANoptosis and multiorgan injury in sepsis. We demonstrate that the TCA cycle intermediate α-KG accumulates in septic immune cells, functioning not as a passive byproduct but as a critical pathological mediator.

Sepsis-associated on immune cell death in sepsis predominantly focuses on inflammatory mediators and PRR ligands, including PAMPs and DAMPs [11, 38]. Emerging evidence classifies certain pro-inflammatory metabolites, such as oxidized low-density lipoproteins, free fatty acids, and advanced glycation end products, as MAMPs, which activate PRR-mediated inflammatory signaling in a manner analogous to DAMPs [39]. This inflammatory cascade is further amplified through reciprocal interactions between PAMPs and PRRs, exacerbating catabolic processes that elevate MAMP concentrations, thereby establishing a self-perpetuating cycle of inflammation and multiorgan dysfunction [40]. Our findings position dysregulated α-KG within this expanding conceptual framework of immunogenic metabolites.

To investigate the pathophysiological role of α-KG in this context, macrophages were subjected to co‐treatment with LPS and DM‐α‐KG. The combinatorial stimulation resulted in a fourfold increase in cell death (40%) compared to DM‐α‐KG treatment alone (10%), indicating a synergistic potentiation of inflammatory lethality. Mechanistically, LPS raised intracellular α-KG accumulation (as verified by metabolomics) while concurrently inducing microenvironmental reprogramming that primes pro-inflammatory pathways. Furthermore, pharmacological inhibition (IDH1-305) or genetic ablation of IDH1 significantly attenuated LPS-induced cytokine secretion (IL-1β, IL-6, TNF-α), confirming α-KG plays a critical role in amplifying inflammatory signaling and regulating the survival-inflammation balance in macrophages during sepsis.

We identified excessive α-KG accumulation as a critical driver of PANoptosis, an inflammatory cell death modality requiring PANoptosome assembly. In synergy with inflammatory signals like LPS, supraphysiological α-KG triggers this coordinated pro-inflammatory cell death in macrophages. To identify the specific sensor responsible for mediating α-KG-induced PANoptosis, we focused on the core components of the PANoptosome. Recent studies have delineated key PANoptosome sensors including AIM2, ZBP1, and NLRP3 [41–43]. Our experimental screening using genetically engineered murine models revealed AIM2 deficiency most effectively attenuated LPS/DM-α-KG-induced PANoptosis in BMDMs. Consistent with in vitro findings, in vivo AIM2 ablation significantly mitigated tissue damage and improved survival rate. These results genetically pinpoint AIM2 within the α-KG-driven pathological cascade. Having identified AIM2 as the key sensor, we next sought to elucidate its specific role in orchestrating PANoptosis beyond its canonical function. While the AIM2 inflammasome has been extensively studied as a cytosolic DNA sensor that triggers pyroptosis [44]. However, its role as a nexus for PANoptosis in sepsis is emerging. Our findings confirm that AIM2 upregulation is sufficient to recruit ASC, NLRP3, and other adaptor proteins into a functional PANoptosome, coordinating simultaneous activation of multiple cell death pathways. This aligns with recent evidence that AIM2 serves as a master regulator of PANoptosis in septic AKI and other organ injuries [45]. The pathological significance of this mechanism is underscored by our observation that AIM2 deletion or inhibition markedly attenuates PANoptosis and organ damage, suggesting that targeting the AIM2-PANoptosome axis could yield broad therapeutic benefits across septic complications.

Mechanistically, α-KG serves as an essential co-substrate for the DNA demethylase TET2, thereby regulating epigenetics, gene expression, and differentiation [46, 47]. α-KG overload enhances TET2-mediated DNA demethylation at the AIM2 promoter, facilitating transcriptional activation. The subsequently elevated AIM2 protein nucleates the assembly of a multiprotein PANoptosome complex, simultaneously activating pyroptotic, apoptotic, and necroptotic executors. Consequently, genetic or pharmacological disruption of IDH1, TET2, or AIM2 effectively blocks this PANoptotic cascade, dampens the cytokine storm, alleviates multi-organ injury, and improves survival in murine septic models. These data establish α-KG-TET2-AIM2 signaling as a metabolic-epigenetic axis governing inflammatory cell death and organ dysfunction.

Our findings reveal a concentration- and context-dependent pathogenic role for α-KG, which contrasts with reports of its anti-inflammatory and protective effects in some contexts of sepsis [25, 48, 49]. This apparent discrepancy can be reconciled by considering the decisive factors of dose and inflammatory context. Previous studies demonstrating protection typically used lower concentrations (2 mM [25]) of standard α-KG. In contrast, we employed 5 mM of the cell-permeable derivative DM-α-KG. It is particularly noteworthy that previous research has demonstrated that high concentrations of DM-α-KG (15 mM) can induce GSDMC-dependent pyroptosis in HeLa cells [29]. This suggests that the concentration window of α-KG is decisive for its function: low to medium concentrations (1–2 mM) may primarily exert anti-inflammatory effects, while higher concentrations (5 mM and above) may trigger inflammatory cell death. This concentration-dependent duality aligns with the behavior of other immunometabolites, most notably itaconate, which exhibits clear concentration-dependent biphasic effects—anti-inflammatory at lower doses and pro-inflammatory/apoptosis-inducing at higher doses [50], particularly under inflammatory priming (e.g., with LPS). Thus, our model positions α-KG not as a universally beneficial or detrimental molecule, but as a context-dependent immunological switch.

Sepsis-induced organ failure results from complex intercellular crosstalk involving diverse immune and parenchymal cells. We have not determined whether this metabolic-epigenetic pathway functions identically in other critical cell types-such as neutrophils, endothelial cells, or parenchymal cells like hepatocytes and renal tubular epithelial cells. It is plausible that α-KG accumulation exerts cell-type-specific effects; for instance, in parenchymal cells, it might directly promote ferroptosis through distinct mechanisms rather than AIM2-mediated PANoptosis [51]. This cellular specificity, if present, delineates the boundary and clinical implication of our current model, which captures only a portion of the axis’s full pathological contribution, potentially underestimating its systemic impact and the therapeutic potential of its inhibition. Additionally, through genetic deletion of IDH1, pharmacological inhibition with IDH1-305, or direct supplementation consistently altered intracellular α-KG levels, we observed a corresponding TET2-dependent regulation of AIM2 and subsequent effected PANoptosis. Furthermore, rescue experiments conducted in AIM2^−/− models, both in vitro and in vivo, confirmed that AIM2 is indispensable for initiating the PANoptosome cascade triggered by the α-KG-TET2 axis. Although TET2 loss-of-function abrogated the phenotype, we cannot rule out potential contributions from other α-KG-dependent enzymes, such as prolyl hydroxylases and histone lysine demethylases (KDMs) [48, 52], which may indirectly modulate inflammatory signaling. To definitively exclude parallel mechanisms, future studies should employ catalytic-inactive TET2 mutants or multi-omics profiling approaches. Finally, clinical translation faces the additional hurdle of the metabolic and immunological heterogeneity during sepsis. While we have defined an approximate plasma α-KG range in septic patients, spatiotemporal fluctuations within tissues could override pathway activity and therapeutic efficacy. High-resolution mapping of α-KG distribution across relevant human compartments is therefore an essential next step toward precision interventions.

In summary, our findings position dysregulated α-KG metabolism as a pivotal link between septic immunometabolism and a uniquely potent form of inflammatory cell death, offering novel mechanistic insights and identifying multiple layers of the IDH1/α-KG/TET2/AIM2 axis as potential therapeutic targets. While this axis is firmly established in macrophages, its broader pathophysiological contribution warrants further investigation.

Conclusions

The present study establishes a pathogenic IDH1/α-KG/TET2/AIM2 axis that links metabolic reprogramming to inflammatory PANoptosis in sepsis. We demonstrate that sepsis upregulates IDH1, leading to the accumulation of α-KG. At supraphysiological levels, α-KG synergizes with PAMPs to exacerbate septic pathology. Mechanistically, α-KG enhances TET2-mediated DNA demethylation of the AIM2 promoter, leading to its transcriptional upregulation. Elevated AIM2 then nucleates the assembly of the PANoptosome, triggering the coordinated activation of pyroptotic, apoptotic, and necroptotic executors-a process termed PANoptosis. pharmacological inhibition of IDH1, depletion of α-KG, blockade of TET2 activity, or ablation of AIM2 function robustly suppresses PANoptosis, curbs systemic inflammation, mitigates multi-organ damage, and improves survival in sepsis. Collectively, our findings identify the IDH1/α-KG/TET2/AIM2 axis as a central pathogenic driver of sepsis-associated immunopathology and highlight its components as mechanistically validated targets for therapeutic intervention in this high-mortality syndrome.

Abbreviations

α-KG

α-Ketoglutarate

TCA

Tricarboxylic acid cycle

PANoptosis

Pyroptosis, Apoptosis, Necroptosisoptosis

AIM2

Absent in melanoma 2

DM-α-KG

Dimethyl-α-ketoglutarate

LPS

Lipopolysaccharide

IDH

Isocitrate dehydrogenase

TET2

Ten-eleven translocation 2

DMOG

Dimethyloxallyl glycine

PAMPs

Pathogen associated molecular patterns

DAMPs

Damage-associated molecular patterns

GSDMC

Gasdermin C

RIPK-3

Receptor-interacting protein kinase 3

ZBP1

Z-DNA-binding protein 1

MEFV

Mediterranean fever ever

NLRP3

NACHT, LRR and PYD domains-containing protein 3

CLP

Cecal ligation and puncture

PBMCs

Peripheral mononuclear macrophages

BMDMs

Bone marrow-derived macrophages

WT

Wild-type

FBS

Fetal bovine serum

PEM

Macrophage

DMEM

Dulbecco's modified eagle's medium

RIPK-1

Receptor-interacting protein kinase 1

pRIPK-1

Phosphorylated receptor-interacting protein kinase 1

MLKL

Mixed lineage kinase domain-like pseudokinase

p-MLKL

Phosphorylated MLKL

ASC

Apoptosis-associated speck-like protein containing a CARD

LDH

Lactate dehydrogenase

ELISA

Enzyme-linked immunosorbent assay

IDH

Isocitrate dehydrogenase

AST

Aspartate aminotransferase

ALT

Alanine aminotransferase

BUN

Blood urea nitrogen

MAMPs

Metabolism-associated molecular patterns

PRRs

Pattern recognition receptors

PCD

Programmed cell death

GSDMD

Gasdermin D

PYD

Pyrin domain

CARD

Caspase activation and recruitment domain

H&E

Hematoxylin and eosin

ChIP

Chromatin immunoprecipitation assays

Authors’ contributions

Jinbao Li, Liangfang Yao, Yun Zou, and Qing Tu: Project administration, Supervision and Funding acquisition. Yi Li, Qing Tu, and Chenchen Liu: Validation, Investigation, and Methodology. Jiamin Ma, Yuwei Chen, and Lin Wang: Data curation and Formal analysis. Jinbao Li, Liangfang Yao, and Yun Zou: Resources. Yi Li and Liangfang Yao: Writing – original draft. Ying Chen, Yuwei Chen and Jiali Zhu: Writing – review and editing.

Funding

This work was supported by the National Natural Science Foundation of China (82272193, 82402520, 82472200, 82202372).

Data availability

All the data were available within the article and its supplementary information files of available from the corresponding author (yao_elina@163.com) upon request corresponding author.

Declarations

Ethics approval and consent to participate

All animal experiments were performed in accordance with the ethical principles of National Institutes of Health guide for the care and use of Laboratory Animals. All studies were reviewed and approved by the Ethics Committee of Shanghai General Hospital Clinical Center (IACUC No. 2023AW062).

Competing interests

The authors declare no competing interests.